example parallel workflow with an rOpenSci package

ropensci_parallel.md

Example parallel workflow

In this example, we use the rgbif package to search for occurrence (lat/long) data for 1000 species

Install and load rgbif

install.packages("rgbif")
library("rgbif")

Get Aves (birds) gbif key

name_backbone("Aves")

Get 1000 bird species names

out <- name_lookup(rank = "Species", higherTaxonKey = 212, limit = 1000)
head(out$data)

Get a vector of keys with species as the names for each key (for your own reference)

keys <- out$data$key
names(keys) <- out$data$canonicalName

Search for occurrences.

Note 1: that the function occ_search accepts many keys so you could pass in all 1000 names, but splittig up in chunks of 250 would allow all to be retrieved faster

Note 2: the default record limit in occ_search is 20 records, so we're getting a max of 20 records back per species, but that can be changed of course

library("plyr")
library("doMC")
registerDoMC(cores = 4)
chunksof10 <- split(keys[1:50], ceiling(seq_along(keys[1:50])/10))
res <- llply(chunksof10, occ_search, hasCoordinate = TRUE, limit = 5, return = 'data', .parallel = TRUE)

Remove species with no data

res2 <- lapply(res, function(x) Filter(is.data.frame, x))

Collapse to a single data.frame and inspect

library("data.table")
df <- setDF(rbindlist(unlist(res2, FALSE), fill = TRUE))

Map a few of the species (mapping 1000 species on a single map would be too much)

library("ggplot2")
gbifmap(df[1:40, ])